INTERACTIVE ART INSTALLATION USING CAMERA AND OBJECT DETECTION

A project using Yolo and Norfair to achieve tracking of individuals (in real time and on video) and informations extraction.

The informations extracted are, realtively to a calibrated zone :

• Inside/Outside of the calibrated zone
• Position in % relative to the calibrated zone
• Moving/Still

This project was realized in the context of the 5 semester project at HEIA-FR, in collaboration with Eikon and Minus3. It was used in an interactive art installation at Murten Licht Festival 2024.

The project was realized under the supervision of Beat Wolf, Damien Goetschi and Jean Hennebert.

Setup

Install necesaries librairies : pip install -r requirements.txt

Execution of the project on a local MP4 file : py main.py -i <RELATIVE_PATH_OF_THE_FILE> -o <OUTPUT_NAME>

Execution of the project on a live input :

• Into main.py, comment io.local_output(frame) et uncomment #io.live_output(frame)

• py main.py -i <CAMERA_IP>:<CAMERA_PORT>/h264Preview_01_main

Manual Calibration

The calibration is made automatically, based on the most luminous area of the input, but you can do it manually.

Into constants of the lines 17 to 27 :

• Modify the value MANUAL_CALIBRATION from False to True

  • Example : MANUAL_CALIBRATION = True

• Modify the value of manual_offset_calibrate_x [offset_left, offset_right].

  • Example : manual_offset_calibrate_x = [[20, 400], [1260, 400]]

• Do the same with the value of manual_offset_calibrate_y [offset_top, offset_down].

  • Example : manual_offset_calibrate_y = [640, 440], [640, 650]